基于容积脉搏波的连续血压测量研究

发布时间：2018-04-17 21:11

本文选题：脉搏波信号 + 脉搏波波群　；参考：《天津理工大学》2017年硕士论文

【摘要】：血压是人体最为重要的生理指标之一,反应了人体心脏和血管的物理及供血功能的状态。在疾病诊断、治疗效果评价、药效评估等方面具有十分重要的指导意义。目前血压主要以单次血压测量为主,仅反映了当时瞬时血压值。从医学角度连续血压的测量会更有利于血压药物服用时间和剂量的指导,从而达到科学降血压的目的。目前医院配备的连续血压测量设备,以间歇式测量为主,通常每隔15~30分钟获得一次血压值,不能达到真正的连续。因此,本文提出了一套推算每个心动周期血压的完整策略,具体研究内容如下:针对双路和多源信号血压测量的设备佩戴复杂的状况,本文采用单路脉搏波进行连续血压测量研究。考虑到病患和老年人群因血管弹性降低造成脉搏波特征点丢失的缺陷,以单次心动周期产生的完整脉搏波波形中的离散点作为推算血压的特征。首先,对采集到的脉搏波数据进行滤波处理,再利用线性逐步回归方法建立血压测量模型,然后,以此模型推算每个心动周期的血压值,最后实现血压的连续测量。通过自采数据与公共数据验证,血压值符合±5%错误率的预期标准。研究中发现,线性逐步回归算法在变量多时收敛较慢,影响血压估算得效率。同时,经医学研究表明,人体血压值得变化是非线性的,线性模型的描述会造成估算结果的低准确度。本文在前面研究的基础上,利用检测人的前期数据采用BP神经网络建立血压测量模型,从而实现血压连续测量。验证实验在自采和MIMIC公共数据库中的脉搏波数据上分别进行了验证。在自采数据上,25名被测者收缩压和舒张压均达到AAMI标准,血压值精度比逐步回归方法有小幅提升。在MIMIC数据库上的33名受试者有大幅提高。同时,实验结果还证明此种方法对于健康年轻人、病患、老年人的血压预测同样有效,符合AAMI标准。
[Abstract]:Blood pressure is one of the most important physiological indexes of human body, which reflects the state of the physical and blood supply function of human heart and blood vessel.It has important guiding significance in disease diagnosis, therapeutic effect evaluation and drug efficacy evaluation.At present, blood pressure is mainly measured by single blood pressure, which only reflects the instantaneous blood pressure at that time.Continuous blood pressure measurement from a medical point of view will be more conducive to the guidance of the time and dosage of blood pressure drugs, so as to achieve the purpose of lowering blood pressure scientifically.At present, the hospital is equipped with continuous blood pressure measurement equipment, mainly intermittent measurement, usually every 1530 minutes to obtain blood pressure values, can not reach the true continuous.Therefore, this paper proposes a complete strategy for estimating blood pressure in each cardiac cycle. The specific contents of the study are as follows: for the equipment with dual and multi-source signal blood pressure measurement,In this paper, single pulse wave is used for continuous blood pressure measurement.Considering the defect of pulse wave characteristic point loss caused by the reduction of vascular elasticity in patients and elderly people, the discrete points in the complete pulse wave waveform produced by a single cardiac cycle are used as the characteristics of estimating blood pressure.Firstly, the pulse wave data collected are filtered, then the blood pressure measurement model is established by linear stepwise regression method. Then, the blood pressure value of each cardiac cycle is calculated by the model, and the continuous blood pressure measurement is realized finally.The self-collected data and common data showed that the blood pressure was up to the expected standard of 卤5% error rate.It is found that the linear stepwise regression algorithm converges slowly in multiple variables, which affects the efficiency of blood pressure estimation.At the same time, medical research shows that the change of human blood pressure is nonlinear, and the description of linear model can result in low accuracy of estimation results.On the basis of the previous research, BP neural network is used to establish the blood pressure measurement model by using the pre-data of human being, so as to realize the continuous blood pressure measurement.The validation experiments were carried out on the pulse wave data of self-mining and MIMIC public database respectively.The systolic and diastolic blood pressure of 25 subjects were up to AAMI standard, and the accuracy of blood pressure was slightly higher than that of stepwise regression method.There was a significant increase in the 33 subjects on the MIMIC database.The results also show that this method is equally effective in predicting blood pressure in healthy young people, patients and the elderly, and conforms to AAMI criteria.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R443.5;TP183

【相似文献】